Google has formally filed a motion [PDF] asking the 9th Circuit Court of Appeals to put a pause on the order that forces the company to open the Play store to competitors. If you’ll recall, Google lost an antitrust lawsuit filed by Epic Games after a federal jury found that the company held an illegal monopoly on app distribution and in-app billing services for Android devices. Earlier this month, US District Judge James Donato ordered Google to allow third-party app stores access to the Google Play app catalog and to make those stores downloadable from its storefront. Now, Google is asking the court for a stay on that order while it’s appealing the Epic antitrust lawsuit decision, saying that it will expose 100 million Android users in the US to “substantial new security risks.”

The company called the order “harmful and unwarranted” and said that if it’s allowed to stand, it will threaten Google’s ability to “provide a safe and trusted used experience.” It argued that if it makes third-party app stores available for download from Google Play, people might think that the company is vouching for them, which could raise “real risks for [its] users.” Those app stores could have “less rigorous protections,” Google explained, that could expose users to harmful and malicious apps.

It also said that giving third-party stores access to the Play catalog could harm businesses that don’t want their products available alongside inappropriate or malicious content. Giving third-party stores access to its entire library could give “bad-intentioned” stores a “veneer of legitimacy.” Moreover, it argued that allowing developers to link out from their apps “creates significant risk of deceptive links,” since bad actors could use the feature for phishing attacks to compromise users’ devices and steal their data.

One of court’s main proposed changes is to allow developers to remove Google Play billing as an option, allowing them to offer their apps to Android users without having to pay the company a commission. However, Google said that by allowing developers to remove its billing system, it could “force an option that may not have the safeguards and features that users expect.”

In its filing, Google emphasized that the three weeks the court gave it to make these sweeping changes is too short for a “Herculean task.” It creates an “unacceptable risk of safety” that could lead to major issues affecting the functionality of users’ Android devices, it said. The company also questioned why the court sided with Epic in its antitrust lawsuit, whereas it sided with Apple in a similar case also filed by the video game company. “It is pause-inducing that Apple, which requires all apps go through its proprietary App Store, is not a monopolist, but Google — which built choice into the Android operating system so device makers can preinstall and users can download competing app stores — was condemned for monopolization.”

Nvidia quietly unveiled a new artificial intelligence model on Tuesday that outperforms offerings from industry leaders OpenAI and Anthropic, marking a significant shift in the company’s AI strategy and potentially reshaping the competitive landscape of the field.

The model, named Llama-3.1-Nemotron-70B-Instruct, appeared on the popular AI platform Hugging Face without fanfare, quickly drawing attention for its exceptional performance across multiple benchmark tests.

Nvidia reports that their new offering achieves top scores in key evaluations, including 85.0 on the Arena Hard benchmark, 57.6 on AlpacaEval 2 LC, and 8.98 on the GPT-4-Turbo MT-Bench.

These scores surpass those of highly regarded models like OpenAI’s GPT-4o and Anthropic’s Claude 3.5 Sonnet, catapulting Nvidia to the forefront of AI language understanding and generation.

Nvidia’s AI gambit: From GPU powerhouse to language model pioneer

This release represents a pivotal moment for Nvidia. Known primarily as the dominant force in graphics processing units (GPUs) that power AI systems, the company now demonstrates its capability to develop sophisticated AI software. This move signals a strategic expansion that could alter the dynamics of the AI industry, challenging the traditional dominance of software-focused companies in large language model development.

Nvidia’s approach to creating Llama-3.1-Nemotron-70B-Instruct involved refining Meta’s open-source Llama 3.1 model using advanced training techniques, including Reinforcement Learning from Human Feedback (RLHF). This method allows the AI to learn from human preferences, potentially leading to more natural and contextually appropriate responses.

With its superior performance, the model has the potential to offer businesses a more capable and cost-efficient alternative to some of the most advanced models on the market.

The model’s ability to handle complex queries without additional prompting or specialized tokens is what sets it apart. In a demonstration, it correctly answered the question “How many r’s are in strawberry?” with a detailed and accurate response, showcasing a nuanced understanding of language and an ability to provide clear explanations.

What makes these results particularly significant is the emphasis on “alignment,” a term in AI research that refers to how well a model’s output matches the needs and preferences of its users. For enterprises, this translates into fewer errors, more helpful responses, and ultimately, better customer satisfaction.

How Nvidia’s new model could reshape business and research

For businesses and organizations exploring AI solutions, Nvidia’s model presents a compelling new option. The company offers free hosted inference through its build.nvidia.com platform, complete with an OpenAI-compatible API interface.

This accessibility makes advanced AI technology more readily available, allowing a broader range of companies to experiment with and implement advanced language models.

The release also highlights a growing shift in the AI landscape toward models that are not only powerful but also customizable. Enterprises today need AI that can be tailored to their specific needs, whether that’s handling customer service inquiries or generating complex reports. Nvidia’s model offers that flexibility, along with top-tier performance, making it a compelling option for businesses across industries.

However, with this power comes responsibility. Like any AI system, Llama-3.1-Nemotron-70B-Instruct is not immune to risks. Nvidia has cautioned that the model has not been tuned for specialized domains like math or legal reasoning, where accuracy is critical. Enterprises will need to ensure they are using the model appropriately and implementing safeguards to prevent errors or misuse.

The AI arms race heats up: Nvidia’s bold move challenges tech giants

Nvidia’s latest model release signals just how fast the AI landscape is shifting. While the long-term impact of Llama-3.1-Nemotron-70B-Instruct remains uncertain, its release marks a clear inflection point in the competition to build the most advanced AI systems.

By moving from hardware into high-performance AI software, Nvidia is forcing other players to reconsider their strategies and accelerate their own R&D. This comes on the heels of the company’s introduction of the NVLM 1.0 family of multimodal models, including the 72-billion-parameter NVLM-D-72B.

These recent releases, particularly the open-source NVLM project, have shown that Nvidia’s AI ambitions go beyond just competing—they are challenging the dominance of proprietary systems like GPT-4o in areas ranging from image interpretation to solving complex problems.

The rapid succession of these releases underscores Nvidia’s ambitious push into AI software development. By offering both multimodal and text-only models that compete with industry leaders, Nvidia is positioning itself as a comprehensive AI solutions provider, leveraging its hardware expertise to create powerful, accessible software tools.

Nvidia’s strategy seems clear: it’s positioning itself as a full-service AI provider, combining its hardware expertise with accessible, high-performance software. This move could reshape the industry, pushing rivals to innovate faster and potentially sparking more open-source collaboration across the field.

As developers test Llama-3.1-Nemotron-70B-Instruct, we’re likely to see new applications emerge across sectors like healthcare, finance, education, and beyond. Its success will ultimately depend on whether it can turn impressive benchmark scores into real-world solutions.

In the coming months, the AI community will closely watch how Llama-3.1-Nemotron-70B-Instruct performs in real-world applications beyond benchmark tests. Its ability to translate high scores into practical, valuable solutions will ultimately determine its long-term impact on the industry and society at large.

Nvidia’s deeper dive into AI model development has intensified the competition. If this is the beginning of a new era in artificial intelligence, it’s one where fully integrated solutions may set the pace for future breakthroughs.

Elias Torres has achieved a lot for somebody who immigrated to the US from Nicaragua at 17 without knowing any English. He served as a VP of engineering at HubSpot before co-founding Drift, a company that sold to Vista Equity for about $1.2 billion in 2021.

“It’s very rare to get this far, but I’m not done,” Torres told TechCrunch.

About a year ago, Torres (pictured above) founded Agency, an AI-powered startup designed to automate tasks traditionally handled by customer success managers (CSMs). These professionals provide personalized support to users of complex B2B software, ranging from onboarding and training to upselling new features.

On Wednesday, Agency is coming out of stealth and announcing that it raised a $12 million seed round led by Sequoia and HubSpot Ventures.

The idea for Agency was born when Torres started consulting for OpenAI in early 2023. The ChatGPT maker asked Torres for help developing AI solutions for some of their enterprise customers, including NBA and LiveNation. In the course of doing that, it occurred to Torres that companies could benefit from AI customer success managers.

He was encouraged to build a startup around this concept when he met with Brian Halligan, co-founder and executive chairman of HubSpot. “We worked together on the CRM at HubSpot, and he told me, ‘Let’s build something great together again,’” Torres said about his conversation with Halligan. (Halligan joined Agency’s board.)

Shortly after meeting with Halligan, Torres reached out to Sequoia partner Pat Grady, who had previously invested in Drift. Grady was instantly sold on the idea.

“It’s hard to hire great CSMs. It’s hard to scale great CSMs,” Grady told TechCrunch. “If you have a product that can do a lot of the work on their behalf, and you can scale your company without having to hire an army of CSMs. That’s pretty useful.”

Agency can free up time in the customer success manager’s workday by handling tasks such as scheduling, follow-ups, note-taking, customer onboarding, and meeting preparation.

Torres explained that Agency’s AI gains a deep understanding of each customer from emails, CRM data, chat messages, and phone conversations, which allows it to anticipate customer needs at any point.

“This is something that we’ve been dreaming about for a long time,” he said, adding that this was the goal of Salesforce, the CRM he helped build at HubSpot and Drift, which was building personalized conversations for salespeople. “We didn’t have the technology to do this until now.”

The company’s product is currently being tested with companies, including HeyGen, and is available in an invite-only beta for customer success professionals.

While Agency seems to have no direct competitors at present, another function within the sales and marketing organization, sales development representative, is facing disruption from dozens of AI-powered solutions.

“I don’t know anybody else going after this market,” Sequoia’s Grady said. “Hopefully people won’t discover that for a while, and they’ll have a little bit of room to run.”

The ruling, which Google has appealed, would force Google to distribute third-party app stores within Google Play, no longer require Google Play Billing for apps distributed via Google Play, and more, with many of those changes ordered to begin on November 1st — just over two weeks from today.

But echoing many of Google’s arguments during the district court case, which Judge Donato rejected as insufficient, the company now argues that the order “threatens Google Play’s ability to provide a safe and trusted user experience.”

“This wouldn’t just hurt Google – this would have negative consequences for Android users, developers and device manufacturers who have built thriving businesses on Android, writes Google’s Lee-Anne Mulholland, VP of regulatory affairs, in a fact sheet distributed to journalists.

The fact sheet is bulleted into five different sections, and the section headers give you an idea of Google’s objections:

To get a sense of Google’s actual filing with the court, here’s how it begins:

At the request of a single competitor, Epic Games, the District Court ordered extensive redesigns to Play that will expose 100-million-plus U.S. users of Android devices to substantial new security risks and force fundamental changes to Google’s contractual and business relationships with hundreds of thousands of Google partners. The court gave Google just three weeks to make many of these sweeping changes—a Herculean task creating an unacceptable risk of safety and security failures within the Android ecosystem.

You can read the whole fact sheet, and Google’s whole emergency motion, below.

ChatGPT became much more collaborative when OpenAI released Canvas mode for the AI chatbot earlier in October. Switching to Canvas mode provides a more flexible way to create and edit text. Through AI’s code-writing ability, it enables more complex, long-term planning with visualization, spot-editing, and even automation.

Despite OpenAI’s bragging about how practical this approach to ChatGPT can be, you might stare at that prompt and ask how to use ChatGPT’s Canvas mode to enhance your daily life.

SpaceX achieved a spectacular first on Sunday when it used a pair of giant mechanical arms to catch the 70-meter-tall Super Heavy booster just minutes after it deployed the Starship spacecraft to orbit in the vehicle’s fifth test flight.

But SpaceX isn’t stopping there. As part of its efforts to create a fully reusable spaceflight system for the Starship — comprising the first-stage Super Heavy booster and the upper-stage Starship spacecraft — SpaceX will attempt to catch not only the booster, but also the spacecraft.

SpaceX CEO Elon Musk confirmed the plan for the world’s most powerful rocket in a post on X ( formally Twitter) on Wednesday, saying, “Hopefully early next year, we will catch the ship too.”

Before then, SpaceX will want to carry out more test flights of the Starship in which it will continue to catch the Super Heavy, while the Starship will continue to come down in the ocean, as it did in Sunday’s test flight.

Catching the Starship back at the launch base will allow for a faster turnaround time between launches, with the spacecraft only needing to be checked, refurbished, and refueled before being lifted atop a Super Heavy for another flight.

SpaceX also has to perfect a landing system for the Starship that involves it touching down on the ground in a vertical position, as this is how it will arrive on other celestial bodies such as the moon and possibly Mars (at least until any launch and landing infrastructure can be built).

It’s actually already achieved such a landing in Earth-based tests several years ago, but those touchdowns involved shorter “hops” into the atmosphere rather than more complex orbital flights.

It’s certainly an exciting time for SpaceX engineers as they put much of their attention into the continued development of the Starship.

NASA is planning to use SpaceX’s spacecraft to put two astronauts on the lunar surface in the Artemis III mission, which is currently scheduled for 2026, so there is much work to be done.